Lakes of digital ink have been spilled on the topic of AI killing traffic to media sites. I’ve certainly poured my share. The basic fear: If your business depends on attracting as many eyeballs as possible to content on a website, AI will detour that gaze and point it toward its own summary of that content, resulting in far fewer people looking your way.
There’s still a lot to be resolved with respect to the economics of AI scraping and how publishers will be compensated for that act. But however that plays out, it’s becoming clearer by the day that the battle for attention is slowly shifting to whose information is cited most prominently in an AI summary. AI presence isn’t a substitute for website traffic, but it’s the new proxy for relevance and authority.
In my first column for Fast Company, I argued that the incentive system this creates is superior to the media environment of a decade ago, when search and social media were the dominant avenues of digital distribution. For more than a decade, media and marketing learned to chase engagement, which led to a fire hose of listicles, outrage bait, and formulaic informational pieces (“What time is the Super Bowl?” et al.). But if AI systems are the arena—and if they really do reward well-sourced, domain-specific content more than social heat—that could lead to a resurgence of good journalism, at least directionally.
Now we have data to help judge whether or not that’s true. AI search engines have been around for well over a year now, and their use is rising fast, so we are starting to understand whether this substance-over-clickbait theory works in practice. And so far it looks like it might, with some significant caveats.
LinkedIn gets the bot bump
A pair of recent studies evaluated data from millions of AI citations—which typically means a source that’s mentioned and linked in an AI summary. The data found that AI systems treat LinkedIn as one of the most authoritative sources on the internet: Research from Meltwater, a communications intelligence company, showed LinkedIn as the second most-cited source overall in AI summaries (after YouTube), and a separate study from Semrush, a search-data analytics company, concurs, also putting it at No. 2, closely behind Reddit.
The Meltwater data also point to why LinkedIn is a decent indicator of substance: Individual members (rather than brand or company accounts) drove most of the citations, structured content—such as newsletters and posts—performed best, and more than half of the citations went to members with fewer than 10,000 followers. Likewise, Semrush found that the most-cited LinkedIn posts had only modest engagement on the network itself. That’s pretty good evidence against a simple popularity model.
However, there’s also a lot of evidence to suggest that AI systems value structure above substance. When you drill down into academic papers that zero in on exactly how large language models prioritize information, like this one from the Canadian AI company Cohere, they show that LLMs will miss key facts when an article lacks clear titles and headings. Another paper from Stanford University shows AI search systems strongly favor the beginning and the end of documents rather than the rest, meaning if the meat of the piece is highlighted only in the middle, it can often get missed.
All this suggests that AI systems are as “gameable” as search engines and social networks, just in a different way. An article that’s optimized for machines—with declarative introductions and conclusions, clear questions and answers, and consistent titling throughout—but otherwise empty of substance has a good chance of being cited over a piece that may have unique and important facts that are mentioned only halfway through. AI systems reward retrievable substance, not necessarily the most insightful or information-dense content.
In other words, simply making your content visible to AI engines isn’t enough; you need to hand-hold bots so they can find the good information within. This of course is the whole idea behind GEO (generative engine optimization), which can sometimes seem anathema to good writing, which features clever titles, hooks, and backing into topics through narratives—all things humans value more than machines.
The human edge in machine search
But the fact that platforms like LinkedIn, YouTube, and Reddit are such highly ranked sources suggests the best content is a mix of machine-friendly formatting and the human element. While it’s true that AI doesn’t always cite the most engaged posts, the Semrush data also shows that frequent posting and having an established following still help. LinkedIn’s own internal guidance points in the same direction. So engagement still matters, just less directly than it did in the previous era.
And demoting raw engagement is progress! As for AI’s bias toward structured content, journalists and content creators can leverage that tendency to work for them. It signals that content based on original reporting or insights needs to do several things: explain concepts clearly and quickly, include machine-friendly structures like subheadings, and connect the dots with other sources the AI is reading by referencing them by name.
The opportunity, then, is to make good work easier for machines to understand without sanding off what made it valuable to humans in the first place. The next incentive system will have its own bad habits, and there’s no doubt many people will try to exploit them. But if AI search gives more weight to original facts, named sources, clear context, and demonstrated expertise than to outrage and raw engagement, that is an opening worth taking seriously. The winners in an AI-mediated future shouldn’t simply be the loudest accounts or the best-formatted posts. They should be the people who know something real and can demonstrate its worth to the right audience—both human and machine.
